This document summarizes research on 400 Gb/s and 1 Tb/s transmission systems and fiber nonlinearities. It discusses: 1. Fiber nonlinear limits on spectral efficiency and distance as a function of launch power, channel count, and linear vs. nonlinear regimes. 2. Experimental results demonstrating a 1 Tb/s super channel over 138 km using 5x224 Gb/s signals with digital signal processing for nonlinear compensation. 3. A field trial of the 1 Tb/s super channel and 112 Gb/s QPSK signals over 138 km and 330 km respectively in the GIGA network in Brazil, showing improved bit error rate with nonlinear compensation.

1. 400 Gb/s & 1 Tb/s systems and
fiber nonlinearities
Jacklyn D. Reis, PhD
Luis H. H. de Carvalho, BSc
Carolina Franciscangelis, BSc
Victor Parahyba, BSc
Júlio C. M. Diniz, MSc
Daniel M. Pataca, PhD
Fábio D. Simões, PhD
Neil G. Gonzalez, PhD
Júlio César R.F. de Oliveira, PhD
CPqD, Campinas, São Paulo, Brazil 23-25 February 2014
Day 1

2. 1. Fiber Nonlinear Limits
2. 1 Tb/s Super Channel
3. Field Trial with Nonlinear Compensation
1. 1 Tb/s super channel 138 km
2. 112 Gb/s DP-QPSK 338 km
Outline

3. Nonlinear Limits in Optical Networking
• Spectral efficiency times distance  [bit/s/Hz]*km
• Longer distance
• Higher launch power
• Higher optical channels
• Linear Regime  ASE
• Nonlinear Regime  Kerr nonlinearities686 JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 28, NO. 4, FEBRUARY 15, 2010
Fig. 35. Spectral efficiency after transmission for various distance. All links
are without dispersion compensation.
impairments. It is possible to avoid this recorrelation of WDM
channels by using dispersion compensators that are “channel-
ized,” i.e., that compensate dispersion independently for each
WDM channel without compensating the relative time delay be-
Fig. 36. Spectral efficiency for four signal and noise scenarios for the 2000 km
transmission of Fig. 35. (ch: channel.)
Essiambre et al, IEEE/OSA JLT’10.
1 2 3 4 5 6 7 8 9 10
100
1000
10000
30000
Distance(km)
Net Spectal Efficiency (b/s/Hz)
1000 (bit×km)/(s×Hz)
4000 (bit×km)/(s×Hz)
10000 (bit×km)/(s×Hz)
40000 (bit×km)/(s×Hz)
Fig. 2. System reach vs. throughput trade-off in main record transmission experiments based
on coherent detection published in the last five years.
gh-order format generation and N-WDM spectral shaping can both be obtained with
l techniques [4, 8], but current consensus strongly favors the use of transmitter (Tx)
Nespola et al, OptEx’14.
Liu et al, IEEE SPMag’14

4. FIBER NONLINEARITIES
PART I

5. Simulation model
• Electrical + optical components for simulation super
channels / WDM systems implemented in Matlab
• DSP  normalization, CD equalizer (2x), LMS (2x),
blind phase search, synchronization, EVM/SNR
ADC
Analog
Filtering
Downsample Quantizer DSP
Jitter
DAC
Symbols Upsample Quantizer
AWGN
Analog
Filtering
ZOH
Upsample
Jitter
ClippingNyquist
ADC
Analog
Filtering
Downsample Quantizer DSP
Jitter
DAC
Symbols Upsample Quantizer
AWGN
Analog
Filtering
ZOH
Upsample
Jitter
ClippingNyquist
Transmitter
λtx
DAC
PBS
DAC
PBC
Mod
Mod
Out
Receiver
DigitalSignalProcessing
Ix
Qx
2x4
90º
Hyb
ADC
ADC
λrx
PBS
In
PBS
Iy
Qy
2x4
90º
Hyb
ADC
ADC
A
1:N
A
W
G
A
W
G

6. Maximum reach without amplification
• If 7% HD FEC with 3.8x10-3 (SNR≈21.1 dB)  Maximum
reach of 175 km with linear compensation only
• Launch power ≈ 4 dBm (single polarization)
Receiver
DSP
Ix
Qx
2x4
90º
Hyb
ADC
ADC
λN
Prx = -31 dBm
Ptx = -31 dBm
+ ILfiber
Transmitter
λ1
DAC
Mod
Out
DAC
A
-21 dBm
-11 dBm
-16 dBm
-6 dBm
-1 dBm
+4 dBm

7. -1 dBmSignal+SPM
SPMSPM
+19 dBm
Maximum reach without amplification
• Output optical spectrum for different input power (or transmission
distance).
• Nonlinear generation from -21 (50 km-SSMF, top left) dBm up to 19 dBm (250
km, bottom right) launch power
-21 dBm-16 dBm-11 dBm-6 dBm+4 dBm+9 dBm+14 dBm

8. Maximum reach without amplification
• Considerations on launch power and receiver sensitivity
• If receiver sensitivity is lower than -33 dBm, than the input power has
to increase
• Input power is upper bounded by fiber nonlinearities (~4 dBm up to
100 km – SSMF)
• To avoid increasing the launch power to support the loss budget,
amplification prior the coherent receiver may be used to enhance
sensitivity

9. Multi-carrier transmission at 50 GHz grid
• Nx43 Gbaud – 64QAM at 50 GHz frequency grid after
100 km of SSMF (0.2 dB/km, 16 ps/nm/km, 1.3 W-1km-
1)
9
−100 −50 0 50 100
−100
−90
−80
−70
−60
−50
−40
−30
−20
−10
0
Output Optical Spectrum
Frequency to 193.4 THz [GHz]
OpticalPower[dBm/Hz]
−100 −50 0 50 100
−100
−90
−80
−70
−60
−50
−40
−30
−20
−10
0
Output Optical Spectrum
Frequency to 193.4 THz [GHz]
OpticalPower[dBm/Hz]
−100 −50 0 50 100
−100
−90
−80
−70
−60
−50
−40
−30
−20
−10
0
Output Optical Spectrum
Frequency to 193.4 THz [GHz]
OpticalPower[dBm/Hz]
−100 −50 0 50 100
−100
−90
−80
−70
−60
−50
−40
−30
−20
−10
0
Frequency to 193.4 THz [GHz]
OpticalPower[dBm/Hz]
Transmitter
Receiver
DSP
Ix
Qx
2x4
90º
Hyb
ADC
ADC
λN
100 km SSMF
0.2 dB/km
16.5 ps/nm/km
1.3 (kmW)-1
A
W
G
Transmitter
TransmitterTransmitter
λ1
DAC
Mod
Out
DAC
A

10. −9 −6 −3 0 3 6
−30
−27.5
−25
−22.5
−20
−17.5
−15
Nch x 43 Gbaud −64 QAM at 100 km −SSMF
Launch Power [dBm]
EVMRMS
[dB]
1 Channel
2 Channels
BER = 10
−3
3 Channels
4 Channels
Multi-carrier transmission at 50 GHz grid
• Intra-channel SPM versus inter-channel XPM at 50 GHz
• From 1 Channel to 4 Channels  3 dB penalty on launch power BER =
10-3 due to XPM

11. 1T SUPER CHANNEL
PART II

12. Optical pre-filtering
• Experimental setup
• WDM scenario: 3 channels, 35-GHz spacing.
• Optical filtering: Nyquist optical filter, bandwidth variation.
• Use of conventional DSP, without ICI or ISI compensation.
• Results analysis
• Measured performance at central channel (worst scenario).
• Q-Penalty in reference to “non-filtered” 224G RZ PDM-16QAM.
• Best performance @ 26GHz bandwidth (ICI vs ISI trade-off).
Test Channel
VOA
2x1
Neighbors
SCOPE(40-GS/s)
(a)
EDFA
3 dB
WaveShaper
Mod. RZ
PDM-16QAM
Mod. RZ
PDM-16QAM
LO
OfflineDSP
(b)
ICR

13. Experimental Setup
Fig. 1: Experimental setup. (a) 1.12-Tb/s transmitter; (b) 224G RZ-PDM-16QAM optical eye and constellations;
(c) 1.12-Tb/s spectrum in a 175-GHz flexi-grid WSS channel; (d) Recirculation loop; (e) DSP-Rx block diagram.

14. Back-to-back characterization (per-carrier)
• Measured performance @ 1e-3 BER
• 224Gb/s RZ PDM-16QAM (Reference): 26 dB (6-dB implementation penalty).
• Reference after filtering: 25.5 dB (0.5-dB matched filter improvement).
• 1.12-Tb/s Superchannel (5-Carriers): 26.3 dB (0.8-dB multiplexing penalty).
• 1.12-Tb/s after 5 ROADMs@175-GHz: 27.4 dB (1.1-dB penalty).
• Required OSNR @ FEC Limit: 3.8e-3 BER
• 1.12-Tb/s Superchannel (5-Carriers): 23.3 dB.

15. Transmission results
Launch-Power after 700km
BER with NLC after 1000 km Transmission performance
OSNR performance (per-carrier) • Launch-Power test
• Per-carrier investigation.
• Optimum value: -1 dBm/carrier
• Transmission results
• 700 km and 7 ROADM passes
• Using conventional DSP
• 1000 km and 10 ROADM
passes
• Employing nonlinear
compensation.
• OSNR performance
• Without NLC
• After 700 km: 25.3 dB
• Transmission penalty: 2 dB
• With NLC
• After 1000 km: 23.9 dB
• Transmission penalty: 0.6 dB
• Improvement of 1.4 dB in
transmission penalty by
employing NLC
• Transmission reach improved
from 700 to 1000 km.

16. FIELD TRIAL
PART III

17. Field Trial: GIGA Network
• Campinas to Jundiaí  ~138 km
• Campinas to São Paulo  ~330 km

18. 1 Tb/s Super Channel: 5 x 224 Gb/s DP-16QAM
RZ+PF
DAC free
6 b/s/Hz SE (175 GHz grid)
ECOC’13

19. −5 −4 −3 −2 −1 0
0
0.002
0.004
0.006
0.008
0.01
0.012
Launch Power per Channel [dBm]
BER
1 Tb/s Super Channel: 5x224 Gb/s DP−16QAM −Campinas −Jundiai
BER = 3.8x10
−3
Asymmetric SSF
1
+MLSE
Symmetric SSF
16
Asymmetric SSF
1
Linear Compensation
1 Tb/s: 5 x 224 Gb/s DP-16QAM  138 km
• BER Improvement at Optimal Power  3.5x

20. 112 Gb/s: DP-QPSK  330 km
• BER Improvement at Optimal Power  1.5x / 2.8x
• Nonlinear tolerance  2 dB improvement w/o MLSE
−8.5 −7.5 −6.5 −5.5 −4.5 −3.5 −2.5 −1.5 −0.5
0
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0.008
0.009
0.01
Launch Power [dBm]
BER
112 Gb/s DP−QPSK −Campinas −Sao Paulo
Linear Compensation
Asymmetric SSF
1
Symmetric SSF
16
Asymmetric SSF
1
+MLSE
BER = 3.8x10
−3

21. Conclusions
• DSP plays a major role on high-speed optical networking
• Fiber nonlinear effects limit the launch power
• Shorter capacity
• Shorter distance
• Digital nonlinear compensation
• At least 2 dB improvement on nonlinear tolerance
• Network parameters in deployed fibers
• Chip implementation is yet to appear

22. Obrigado!
www.cpqd.com.br